A.D. Alobaidani, * D. Furniss, M.S. Johnson, Faculty of Engineering, University of Nottingham,

Transcription

1 OPTIMISATION AND SPECTROSCOPIC EXPERIMENTS OF CATIONIC EPOXY FORMULATIONS FOR THE APPLICATION OF PHOTOCURING OF DEEP RESIN MEDIUMS USING SIDE EMITTING OPTICAL FIBRES A.D. Alobaidani, * D. Furniss, M.S. Johnson, Faculty of Engineering, University of Nottingham, * Wolfson Centre for Material Research University Park, Nottingham, NG7 2RD, U.K. eaxad@nottingham.ac.uk SUMMARY The photocuring behaviour of several cationic epoxy resins is investigated by curing thick components. Varying the compositional ratio of the epoxy resin formulation resulted in significantly faster curing of a 5 mm thick laminate compared to other commercially available epoxy resins used for adhesive applications. The optimised epoxy based formulations showed low absorption spectrum. Photosensitizers reduce the curing efficiency of thick components. KEYWORDS: epoxy resin, photo-polymerisation, emission spectrum, photosensitizer, oxetane, ultraviolet radiation curing, radiation absorption INTRODUCTION Radiation curing of cationic epoxy formulations is an alternative to thermal curing. It offers a number of advantages compared to thermally cured epoxy, for the following reasons [1]: the reactions are solvent free, there are no residual amines, and the reaction can be carried out at low temperatures. In photoinitiated curing of cationic epoxy, two base epoxy resins are used; diglycidyl ether and cycloaliphatic epoxy resins. These types of epoxy have been studied extensively by Crivello et al. [2], Decker et al. [3] and other authors [4-7] Typically, for a bonded structure, the formulation is encapsulated by opaque substrates preventing activation of the photoinitiator by a conventional UV radiation lamp. Connecting an UV radiation source to a side emitting optical fibre which is implanted within the bond line allows direct activation of the photoinitiator. The optical transmission of the side emitting optical fibres in UV radiation and enhancements to the PMMA optical fibres for efficient curing results were earlier reported [8]. For an efficient photocuring process, the radiation absorption spectrum of a resin system and the emission spectrum of a radiation source should be matched [3, 9]. To photocure thick components the ratio of the photoinitiator in a formulation has to be adjusted to allow the lamp emission to penetrate deep into resin mediums [10]. During the photocuring process, the total rate of energy absorption [8] per unit length (depth of resin) by the photoinitiator to produce reactive species is given by:

2 e E 1 0 e k l d, where, is the wavelength (nm), is the response spectrum (cm -1 ) of the resin system and E is the emission spectrum (W/cm 2 ) of the radiation source. The response spectrum can be determined using the Beer-Lambert law, and hence, it is described as follows: k ln 10 l OD, where, OD is the optical density and l is the path length (cm) through which the radiation travels. The total applied energy [11] of radiation delivered by a source to cure a resin formulation may be calculated by: Q = E t where, Q is the total applied energy (J/cm 2 ), E is the total delivered irradiance (W/cm 2 ), t is the duration of illumination (s). This study investigates the deep curing characteristics of several types of epoxy resin systems: photo-polymerisation speed and optical transmission before as well as during the photocuring process in the spectral range from UV-A to visible blue. MATERIALS AND METHODS The characterisation process was carried out for two commercial pre-formulated epoxy resin ( and , Dymax Corp.) and four epoxy based formulations prepared in-house using two types of base epoxy resins: bisphenol A/F purchased from Dow Chemical Company and cycloaliphatic epoxy (UVR-6105) purchased from Univar Europe. The in-house prepared formulations using the base epoxy resins were made using triarylsulfonium hexafluorophosphate salts (UVI-6992) and triarylsulfonium hexafluoroantimonate salts (UVI-6976) photoinitiators, both purchased from Dow Chemical Company. Oxetane (OXT-101) polyol was also used to enhance the reactivity and the mechanical properties. The formulation prepared using the cycloaliphatic epoxy (UVR-6105) was further studied using photosensitizers (anthracene, Sigma-Aldrich Company Ltd, and H-Nu470, Spectra Group Ltd.). A 2000-EC flood lamp (Dymax Corp.) was used for the resin characterisation process. The lamp is equipped with a mercury arc metal halide D-type iron doped bulb that delivers 0.1 W/cm 2 out-put over the spectral range from 280 nm to 450 nm. Maximum efficiency occurs after seven minutes (warm-up). The generated temperature by

3 irradiation on the top side of the casting mould was 73 C (measured using a thermocouple). The hardness was measured using a Barcol Hardness Tester. The curing behaviour of epoxy resin was studied by curing thick samples. A casting mould consisting of an aluminium form with top and bottom toughened glass covers was designed (Figure 1). The toughened glass covers, with a thickness of 7 mm, allowed radiation penetration into the resin. The top glass cover absorbed ~ 6 % of the total emitted radiation. The mould was used for casting 150 x 110 x 5 mm samples for studying the curing behaviour of the different resin systems. Figure 1. Schematic diagram showing the mould parts. Figure 2 illustrates the positioning of the equipment used for finding the absorption spectrum and absorption spectrum variation during the photocuring of the resin systems. The absorption spectrum of the resin systems and the radiation absorbed by the resin systems during the photocuring process were conducted using a 60 watt tungsten halogen lamp. The lamp emits intensity continuously over a range from 300 to 1000 nm. This lamp was tested to determine whether if would photocure the resin systems by a continuous exposure time of ten minutes. All samples indicated no curing. Thus, it was kept on through all the experiments. For the photocuring process of the resin systems, an OmniCure S2000 (EXFO Corp.) radiation source with a nominal flux density (radiant power per unit area) of 30 W/cm 2 over the spectral range from 280 to 450 nm was used. This radiation source was designed to deliver a light beam for spot curing applications. It is equipped with a high pressure mercury vapor bulb with major emission peaks at wavelengths of 365 nm, 406 nm and 440 nm. A reflector within the radiation source is set-up to deliver a light beam through two silica light guide cables with a diameter of 3 mm and length of 1000 mm (Figure 2). The spectral outputs of all the experiments were measured using a USB4000 spectrometer (Ocean Optics). The spectrometer determined the absorbance spectra which indicate how much light a sample absorbs, in the range from 200 nm to 850 nm. The radiation was collected using a 1 mm diameter optical fibre (Figure 2) with an acceptance angle ac of Data processing of the absorption readings (OD) were

4 conducted using software provided with USB4000 spectrometer. The transmission spectral outputs through the epoxy formulations were compared with an emission spectrum of the mercury (Hg) lamp (Dymax Corp.) which is intended to be used for the curing process of an optimised epoxy formulation using side emitting optical fibres. Figure 2. Top view photographic image showing the positioning of the equipment setup used for all of the absorption measurements. The tungsten halogen lamp emission point is 80 mm away from the sample. The silica fibres of the S2000 lamp are 50 mm away. RESULTS AND DISCUSSION Optimisation Process of Different Epoxy Resin Systems The commercial epoxy pre-formulated formulation was cured as received. The ratio of the in-house formulated resin systems from epoxy base resins (bisphenol A/F and cycloaliphatic epoxy) were determined from a repeated curing process by varying the ratios of the sulfonium salts (UVI-6992 and UVI-6976) photoinitiators and the oxetane polyol. The resulting resin systems are shown in Table 1. The polymerisation of these resin system was tested via the hardness (BS [12]) of the top and bottom surfaces of the castings. Efficient photocuring results for the base epoxy resins bisphenol A/F and cycloaliphatic epoxy were reached when UVI-6992 and UVI-6976 photoinitiators, respectively, were used with each base epoxy resin. Table 1. Optimised resin systems of various formulations from the characterisation process. The compositional ratios are described by parts per hundred resin (PHR) Material RS1 RS2 RS3 RS4 Epoxy (Bisphenol A/F) Cycloaliphatic epoxy (UVR-6105) UVI-6992 photoinitiator UVI-6976 photoinitiator Oxetane additive (OXT 101) Anthracene photosensitizer

5 The commercial epoxy resins cured in 65 minutes, while the in-house prepared epoxy formulations cured in 12 minutes. The experimental results of the optimised epoxy formulations are shown in Table 2. The effect of viscosity reduction by the oxetane additive had an influence on both the cycloaliphatic epoxy and the bisphenol A/F. The optimised resin system from bisphenol A/F epoxy had a 48 % reduction in viscosity (186 mpa s) compared to the base resin (355 mpa s). The optimised resin system from cycloaliphatic epoxy was 14 % less viscous (252 mpa s) than the base resin (292 mpa s). The photocured samples from formulations prepared using cycloaliphatic epoxy exhibited a more uniform hardness distribution compared to the photocured samples using bisphenol A/F (Table 2) and the commercial pre-formulated epoxy resins. For further improvements in the reactivity of the resin formulations, photosensitizers were used. The resin formulation containing cycloaliphatic epoxy with the photosensitizer H-Nu470 cured only from the surface facing the flood lamp. The bottom surface remained liquid. Hence, the irradiation was blocked from penetrating to the lower surface as a result of the developed optical property of the cured layer facing the flood lamp. The resin system with the anthracene photosensitizers (RS4) cured in a period 6 times longer than resin systems RS2 and RS3. As resin systems RS2 and RS3 cured faster than the other resins, they consumed the lowest delivered radiation energy Q (Table 2). Thus, resin systems RS2 and RS3 are the most efficient resins amongst all the resin systems. Table 2. Characterisation results of the optimised epoxy resin systems (Refer to Table 1 for the compositional ratio of the resin systems listed in this Table). Resin system Viscosity (mpa s) Curing time (minutes) Average hardness Q (J/cm 2 ) Top Bottom RS RS RS RS Optical Properties of the Optimised Epoxy Resin Systems in a Liquid State and During Photo-polymerisation (Spectroscopic Experiments) The optical transmission properties of the resin systems (Table 1) were investigated before and during the photo-polymerisation process. These properties were determined via their absorption band by monitoring the spectral change of the tungsten halogen lamp emission through the resin samples (5 mm thick). For the optical transmission measurement of the liquid state resins, the absorption results of formulations RS1, RS2 and RS3 showed a variation in the response spectrum λ for each resin system (Figure 3). The resulting response spectrums of all these three resin systems covered some of the Hg lamp emission peak at wavelength 368 nm. Generally, all of the response spectrums λ lay in the 300 nm to 400 nm spectral range and the difference between the response spectrums λ is in the right side of the

6 spectrums. Resin system RS1of bisphenol A/F epoxy shows a higher response spectrum than resin systems RS2 and RS3 of cycloaliphatic epoxy. Resin system RS1 also covers most of the major emission peak at wavelength 368 nm of the Hg lamp. Resin system RS3 has a higher absorption response of about 1 cm -1 at wavelength 353 nm than resin system RS2. This could be explained as a result of the oxetane additive in resin system RS3. Resin system RS4, which is a composition of resin system RS3 with anthracene photosensitizer, has the highest response spectrum among all the resin systems. Thus, adding anthracene photosensitizer to resin system RS3 improves its absorption. The response spectrum λ of RS4 overlaps the entire major peak at wavelength 368 nm of the Hg lamp emission. Figure 3. Emission spectrum E λ of Hg lamp and response spectrums λ of Resin systems: RS1, RS2, RS3, and RS4. (Refer to Table 1 for resins ratios). During the photocuring process of resin RS1, the response spectrum 368 increased gradually in the first two minutes of irradiation exposure (Figure 4a); from 3.48 cm -1 to 5.69 cm -1. The sample fully cured in the fourth minute and the measured 368 value was 5.64 cm -1. The response spectrum 368 then decreased in small rates after the fourth minute until it reached 368 of about 5.54 cm -1 by the end of the experiment. The response spectrum 368 of RS2 increased rapidly from 0.5 cm -1 until it reached a maximum of 3.85 cm -1 at the first half minute of the curing process (Figure 4b). The response spectrum 368 then decreased steadily to 2.62 cm -1 after one minute and forty seconds at which the sample was fully cured. The response spectrum 368 increased slightly to 2.74 cm -1 by the end of the experiment. In case of resin system RS3, the response spectrum 368 also increased rapidly from 0.5 cm -1 until a maximum of 2.8cm -1 after 10 seconds of the curing process (Figure 4c). The response spectrum 368 thereafter dropped suddenly in the next 40 seconds to 1.53 cm -1, by which time the sample was fully cured. The response spectrum 368 of the cured sample of resin RS3 showed a constant behaviour at 1.5 cm -1 for the rest of irradiation process.

7 The overall change in the response spectrum 368 of resin system RS4 was very small, as shown in the response spectrum axis of Figure 4d. The response spectrum 368 readings in the first four minutes of the curing process resulted in noise; hence, the reading was ignored. The change in the response spectrum 368 readings caused by curing started from the fourth minute to the ninth minute by a small, gradual increase of 368 from 5.11 cm -1 to 5.28 cm -1, respectively. The response spectrum 368 dropped after that to 5.25 cm -1 in the tenth minute. The sample was fully cured at the ninth minute. Figure 4.Change in response spectrum 368 during the curing of 5 mm thick Resin systems: (a) RS1, (b) RS2, (c) RS3 and (d) RS4. (Refer to Table 1 for resins ratios). All the cured samples indicated higher absorption of wavelength 368 nm than the initial reading of the base uncured resin formulation. Samples cured from resin systems RS2 and RS3 showed a faster curing rate (within 2 minutes) than the samples cured from resin systems RS1 and RS4 which cured in 4 minutes. However, the cured samples of RS2 and RS3 exhibited higher internal stress than the cured samples of RS1 and RS4 as can be observed from the crack in Figure 5. The response spectrum 368 of the cured sample of RS1, with epoxy Bisphenol A/F, decreased slightly after the sample was cured. This indicated the high absorption of the cured sample compared to the resin before curing. The slow change in the response spectrum of the cured sample of RS1 is also related to its yellow colour which was darker than the other samples [13]. The polymerisation of resin systems RS2 and RS3 were faster than resin system RS1, which is in agreement with the literature due to the higher reactivity of cycloaliphatic epoxy resins compared to the DGEBA epoxy resins [14]. The addition of oxetane in RS3 resulted in more transmission of light (Figures 4b and 4c) through the cured sample compared to the other samples. The cured sample of RS3 also exhibited less yellow colour than RS1, RS2 and RS4 (Figure 5). Hence, resin system RS3 allowed more radiation to go through it as it cures compared to all the other resin system. Generally,

8 the yellow colour of the cured epoxy samples increases the radiation absorption in the UVA band as reported by Asilturk et al. [13] which may explain the higher 368 values of the cured samples of the different resin formulations compared to the initial 368 values, as shown in Figures 4a to 4c. The sample of resin system RS4 was expected to have the fastest curing behaviour according to the literature [4, 15]. In addition, RS4 in its initial liquid state had a response spectrum that overlapped the entire major emission peak at wavelength 368 nm of the Hg lamp (Figure 3). But in contrast, among all of the resin systems, resin system RS4 took a longer time to cure. The casting of resin system RS4 took 6 times longer to cure than RS2 and RS3, as shown in Table 2. Although, the cured sample of RS4 exhibited the lightest yellowing colour (Figure 5), it had high absorption after curing (Figure 4d). This indicates that the photosensitizer is still causing high radiation absorption even after the full curing of the sample. One explanation is that, the sample surface facing the radiation source was absorbing highly and limited the irradiation intensity to penetrate efficiently to cure deeper into the medium. Figure 5. Cured samples of: RS1, RS2, RS3 and RS4. (Refer to Table 1 for resins ratio). Interpretation of the Overall Epoxy Optimisation Results Photocuring efficiency of different epoxy resin systems, during the characterisation process, resulted in a different curing behaviour. Generally, resin system RS1 demonstrated improved curing results, such as curing speed and overall hardness, compared to the two commercial pre-formulated epoxy resin systems. However, the variation in the hardness of the top and the bottom surfaces (Table 2) of the casting from resin system RS1 may have resulted from the continuous high response spectrum 368 (5.69 cm -1 at 368 nm wavelength) of the RS1 starting from the second minute of irradiation (Figure 4a). Thus, the hardness of the casting cured from RS1 is irregularly distributed across its surface. This absorption disadvantage not only varied the hardness of the top and bottom surfaces, it also increased the curing period. Nevertheless, adding oxetane in bisphenol A/F epoxy resin system increased the resin system (RS1) conversion speed during the photocuring process. Resin system RS4 exhibited the lowest viscosity compared to the other two resin systems made from cycloaliphatic epoxy. As it is based upon the most reactive resin system (RS3) it was expected to be more reactive. However, RS4 took 6 times longer to

9 cure than resin systems RS2 and RS3. Hence, the curing behaviour result of RS4 contradicts the literature [4, 15] and suggests that anthracene and H-Nu70 photosensitizers are more applicable in formulations used for curing thin films, rather than for thick laminates. In addition, the casting made from resin system RS4 showed improved hardness results compared to resin system RS1. However, it is less efficient than RS1 in terms of curing speed and energy consumption. Unlike resin systems RS1 and RS4, resin systems RS2 and RS3 have more efficient curing behaviour. These resin systems cured in 2 minutes, consumed a total energy Q of 12 J/cm 2 and exhibited uniform hardness distribution through the casting (Table 2). The added oxetane ratio in resin system RS3 reduced its viscosity, and also produced a lower response spectrum 368 of the casting after curing compared to RS2 (Figures 4b and 4c), 1.50 cm -1 and 2.74 cm -1, respectively. This improvement allows the irradiation emission at wavelength 368 nm to penetrate deeper in RS3 making it more suitable for thick components. The lower viscosity of RS3 compared to RS2 (Table 2) also makes it more advantageous for closed mould processes, like Resin Transfer Moulding. Thus, resin system RS3 is the most efficient option among the other optimised epoxy resin systems to be photocured by irradiation emitted from the side of an embedded side emitting optical fibre. CONCLUSIONS Photo-polymerisation and the optical transmission properties of different epoxy formulations (5 mm thick) were investigated. The epoxy formulation optimisation process resulted in four resin systems. Three of these were based on cycloaliphatic epoxy and one on bisphenol A/F epoxy. In-house formulated resin systems from bisphenol A/F and cycloaliphatic epoxides cured faster ( 12 minutes) than the commercially pre-formulated epoxy resin systems (65 minutes). Amongst all the epoxy formulations, the castings from the cycloaliphatic epoxides exhibited more uniformly distributed hardness. Adding oxetane to the bisphenol A/F epoxy formulation improved the curing speed and the overall hardness distribution of the casting. The Barcol hardness of the casting from bisphenol A/F epoxy with oxetane improved by 12 % on the top surface (78) and 14 % on the bottom surface (64) compared to the formulation without oxetane. The castings produced from resin systems of cycloaliphatic epoxy were found to be harder (89 Barcol hardness) with a more uniform hardness distribution than those produced from the resin system of bisphenol A/F epoxy. Resin systems based on cycloaliphatic epoxy consumed a lower energy (12 J/cm 2 ) than the bisphenol A/F based resin system (60 J/cm 2 ). H-Nu70 photosensitizers resulted in poor curing for castings from cycloaliphatic epoxy resin, and anthracene photosensitizers slowed the curing process by 10 minutes when added to the optimised cycloaliphatic epoxy resin system which cured in 2 minutes. The response spectrum λ of all of the optimised resin systems ranged from 300 nm to 395 nm. Albeit, the response spectrum λ of the optimised resin systems varied. None of the optimised resin systems showed complete transparency of the wavelength 368 nm (at which the highest intensity is emitted by the S2000 lamp in the UVA band) after curing. The absorption of the cured samples from the optimised resin systems varied. The optimised resin systems based on bisphenol A/F and cycloaliphatic epoxy with the anthracene photosensitizer cured in a longer time as a result of high absorption during curing ( λ at wavelength 368 nm, 5.3 cm -1 and 5.28 cm -1 ; respectively). The two resin

10 systems of cycloaliphatic epoxy with photoinitiator UVI-6976 (0.6 PHR) and either with or without oxetane additive cured faster as a result of a decrease in absorption during curing ( λ at wavelength 368 nm, 2.74 cm -1 and 1.50 cm -1, respectively). Thus, resin system RS3 is the most efficient option to be photocured by irradiation emitted from the side of an embedded side emitting optical fibre. REFERENCES 1.Nowers J.R. Narasimhan B. The effect of interpenetrating polymer network formation on polymerization kinetics in an epoxy-acrylate system. Polymer, 47, 2006, pp Decker C. Photoinitiated Crosslinking Polymerization. Progress in Polymer Science, 21, 1996, pp Crivello J.V. Photoinitiated Cationic Polymerization. Annual Review of Material Science, 13, 1983, pp Cho J.D, Kim E.O, Kim H.K, Hong J.W. An investigation of the surface properties and curing behaviour of photocurable cationic films photosensitized by anthracene. Polymer Testing, 21, 2002, pp Yagci Y, Schnabel W. On the mechanism of photoinitiated cationic polymerization in the presence of polyols. Die Angewandte Makromolekulare Chemie, 270, 1999, pp Sangermano M, Malucelli G, Bongiovanni R, Priola A. Photopolymerization of oxetane based systems. European Polymer Journal, 40, 2004, pp Olsson R.T, Bair H.E, Kuck V, Hale A. Acceleration of the cationic polymerization of an epoxy with hexanediol. Journal of Thermal Analysis and Calorimetry, 76, 2004, pp Alobaidani A.D, Furniss D, Endruweit A, Johnson M.S, Benson T, Seddon A.B. Enhancement of the side emission efficiency of commercial PMMA optical fibres in the UV-A and visible blue spectrum for photocuring of epoxy resins. 13 th European Conference on Composite Materials (ECCM13). June 2-5, 2008 Stockholm, Sweden. 9.Stowe R.W. Key factors in the UV curing process-the relationship of exposure conditions and measurement in UV process design and process control: Part I- Introduction. Fusion UV System Inc., Gaithersburg, Md., April 2002, pp Narayanan V, Scranton A.B. Photopolymerization of composites. Trends in Polymer Science, 1997, 5, pp Chen Y, Ferracane J.L, Prahl S.A. A pilot study of a simple photon migration model for predicting depth of cure in dental composite. Dental Materials, 21, 2005, pp BS : Method 1001: 1977 EN 59, Measurement of hardness by means of a Barcol impressor. British standard. 13. Asilturk M, Sayılkan F, Sayılkan H, Icduygu G. The synthesis and application of Pb-doped GLYMO/Chelated-Zirconium complex coating materials for UV light absorption. Journal of Applied Polymer Science, 99, 2006, pp Crivello J.V. UV and electron beam-induced cationic polymerization. Nuclear Instruments and Methods in Physics Research B, 151, 1999, pp Crivello J.V, Jang M. Anthracene electron-transfer photosensitizers for onium salt induced cationic polymerizations. Journal of Photochemistry and Photobiology A: Chemistry, 159, 2003, pp